Qwen/Qwen3.6-35B-A3B

Timeline

1. Oct 19, 2026 · Opinion / analysis · 1 source
   Kapitsa: open-source Apple Silicon optimized Metal engine built for Qwen and Nemotron. See benchmarks of Qwen 27B, Nemotron + 8 other models on an M1 Pro (16GB), M4 Pro (48GB) and M5 Max (128GB)
   Kapitsa runs only two model families: Qwen (3.5 4B, 3.8 27B, 3.6 35B-A3B, and 3.8 Flash-Next on 128 GB Macs) and NVIDIA Nemotron (3.5 Lightning, Bolt and Super-VL, plus Dory 4B (this is new, we didnt test Dory too much yet). focusing on only these models allows us customize the Metal kernels for each one on Apple Silicon
2. Oct 19, 2026 · Model release · 2 sources
   Nemotron 3.5 Bolt, Muse Spark Open, and a Fast LFM2.5 24B
   Nemotron 3.5, with a closer look at Bolt and its “native 2-bit” quantization
3. Oct 18, 2026 · Opinion / analysis · 1 source
   I messed up. Buying Z9PE-D8 WS with 4 x16 Slots and 256GB DDR3 RDIMM
   I looked at this motherboard with 256 RDIMMs and the x16 slots.
4. Oct 18, 2026 · Opinion / analysis · 1 source
   Nine models, five coding tasks, one 16GB GPU
   I ran five coding tasks against nine models on one RX 9060 XT (16 GB).
5. Oct 16, 2026 · Opinion / analysis · 1 source
   Understanding How Neural Networks Learn (with interactive visualizations)
   This essay rewrites fundamental assumptions about how neural networks learn and connect, with some of the best interactive visualizations I've ever seen.
6. Oct 15, 2026 · Opinion / analysis · 1 source
   What a £300, 35W Snapdragon mini PC can actually do 👀: 35B coder at ~30 tok/s, and an 80B model that doesnt even fit in RAM running at 17 tok/s
   £299.99 Lenovo IdeaCentre Mini (Snapdragon X1-26-100, 32GB, CPU only):
7. Oct 15, 2026 · Research paper · 2 sources
   Data World: Scaling Data Environments from Scratch
   We introduce Data World, a multi-agent engine that synthesizes environments, tasks, and rubrics from explicit mechanisms.
8. Oct 14, 2026 · Opinion / analysis · 1 source
   cuda, hip: GDN chunked kernel by am17an · Pull Request #29353 · ggml-org/llama.cpp
   Performance on Model Prompt tokens master tok/s pr tok/s Change
9. Oct 14, 2026 · Model release · 3 sources
   Run fast, private agents on-device with LFM2.5-24B-A2B
   Best-in-class agent: Competitive with Qwen3.6-35B-A3B, a model with ~50% more total and active parameters, on agentic and tool-use tasks, scoring 48.09 on LiveClaw versus 38.19.
10. Oct 14, 2026 · Research paper · 1 source
    Are You Reading the Right Thing? Rethinking Internal Signals for Prompt-Injection Detection
    Several recent studies propose detecting prompt injections from a language model's internal state.
11. Oct 13, 2026 · Research paper · 1 source
    AssetForge: Asset-Augmented Task Synthesis for Business Automation Agents
    We introduce AssetForge, an asset-augmented task-construction pipeline that pairs a reusable construction-asset library with task-specific Rubrics.
12. Oct 12, 2026 · Opinion / analysis · 1 source
    llama.cpp optimizations for specific hardware and models
    I've made a few posts recently, just talking about the hardware I've got and the model I use.
13. Oct 12, 2026 · Opinion / analysis · 1 source
    Qwen3.8-27B vs Frontier vs Others
    Here're results from my final test of coding, coding (panda/numpy), data analysis, data-driven decision-making, and local RAG domains I interact with most often.
14. Oct 12, 2026 · Model release · 1 source
    openbmb/MiniCPM-V-4.7-35B-A3B
    With our Lookahead Expert Scheduling (LES) technique introduced alongside MiniCPM-V 4.7, combined with fine-grained layer-wise quantization and DSpark speculative sampling integration, MiniCPM-V 4.7 35B-A3B becomes the first open-source MoE model to run smoothly on-device without compromising performance, achieving a decoding speed of 13+ tokens/s on iPhone 17 Pro Max.
15. Oct 12, 2026 · Opinion / analysis · 1 source
    strix-llama new branch: prompt processing +21–54% over Vulkan and up to 2.7× over stock ROCm kernel at 128k, same outputs quality
    I've been tuning llama.cpp on my Strix Halo box (Radeon 8060S, 128 GB) and the numbers are good enough to share.
16. Oct 11, 2026 · Tutorial / explainer · 1 source
    Running Next Flash IQ3_XXS at ~70 tok/s with 100k context or 2 instances of Qwen 3.6 35B A3B IQ4 at ~145 tok/s with 256k all on $500 of ex mining BC-250 boards
    This will be my third update on the bc-250 cluster and for my first forray into local ai I have been having a blast.
17. Oct 10, 2026 · Research paper · 2 sources
    Vela: User Behavior as Supervision Sparks LLM Intelligence
    We propose user behavior as supervision and introduce Vela, a framework that converts real user behavior data into trainable tasks and trajectories.
18. Oct 9, 2026 · Research paper · 1 source
    Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
    We present KnowForge, a pipeline separated by job: a hybrid-retrieval knowledge-point index; a Locator agent that can only search; a code-scheduled bridge judge that forces the missing second hop; a typed, size-bounded evidence pool that merges every search's ranking; a Deep Reader agent that can only read text, images and tables; and a synthesizer with a format checker.
19. Oct 9, 2026 · Research paper · 1 source
    AutoEnvScaling: Automating the Data Flywheel with Terminal Agents
    We introduce AutoEnvScaling, which treats environment design as a terminal task and uses agents to automate the data flywheel.
20. Sep 28, 2026 · Model release · 1 source
    Holo4: powering generalist computer-use agents
    Holo4 is our new series of agentic models.

Often appears with